Greek language pack (grc) — polytonic diacritic stripping + orthographic normalization

[bsesic]

Summary

Add a Greek language pack (grc) to TRACE, mirroring the structure of the Arabic (ara, #16) and Hebrew (hbo) packs. Second classical pack; unlocks LXX, New Testament, and patristic material.

Motivation

Greek is a planned roadmap pack (LXX, NT, patristic) and a prerequisite for the cross-tradition / Hexapla-style work (roadmap Stage 6). It reuses the language-pack architecture proven by the Arabic pack.

Scope

Mirror src/tracealign/lang/arabic/:

• Normalization (normalize.py): strip polytonic diacritics (acute/grave/circumflex accents, smooth/rough breathings, iota subscript, diaeresis) as a representations skeleton — via NFD + combining-mark (Mn) removal. Fold final sigma ς → σ; lowercase; optional lunate-sigma handling. Keep the diplomatic form in Token.raw; text = diacritic-stripped.
• Tokenization (tokenize.py): Greek has no Arabic-style proclitics; the main concern is elision (apostrophe) and possibly enclitics. Start conservative — likely a near-identity post_tokenize; document what is deferred.
• Scoring tiers (scoring.py): EXACT → DIACRITICS_STRIPPED → ORTHOGRAPHIC_VARIANT (sigma/case folding) → fuzzy, reusing the shared Reason enum values already added for Arabic.

Acceptance criteria

• list_languages() includes grc; tokenize/align with lang="grc" work end-to-end.
• Polytonic → monotonic folding collapses accent/breathing variants into the skeleton; final sigma normalized; diplomatic form preserved in raw.
• Tiered scoring returns reason tags from the existing Reason enum (extend only if a genuinely Greek-specific reason is justified).
• Tests follow TDD; full suite green on 3.10/3.11/3.12; flake8 clean.
• pack.version set (grc-0.1.0) and surfaced in result params.

Notes

• Rule-based and dependency-light (stdlib + existing rapidfuzz), consistent with the project ethos. No new third-party dependencies.
• Roadmap: planned non-Hebrew language packs (Aramaic/Greek/Latin/Arabic/Persian).